A Spatially-explicit, Hierarchical Model of Pattern-process Dynamics of Grassland Landscape (Jianguo Wu)

本研究的目标是：(1)建立一个系统的、计算机化的程序来检测贾斯珀岭草地景观的结构模式，主要基于二维网络函数插值理论；(2)基于碧玉岭草地生态系统数据，构建并验证分层结构、空间显式、随机的模式过程动力学模拟模型。该模型将考虑涉及小尺度/高频、中尺度/低频和大尺度/甚低频现象的过程。每个过程中的随机性都必须考虑在内。大尺度现象可能包括天气波动（如降水、温度）、风/水侵蚀、长期干旱和时间不规律的火灾。另一方面，小尺度的现象可能涉及物理环境的斑块和造丘动物（如地鼠和蚂蚁）的活动。总体而言，该模型将纳入草地小尺度扰动的人口统计学特征，监测斑块的时间变化，并在不同尺度上追踪斑块的老化和灭绝。空间显式的多尺度模式归纳模型采用自下而上的方法，从层次分解的较低层次的动态中导出草地的景观级动态。模式的时空统计分析的计算需求，以及模型的灵敏度分析和蒙特卡罗模拟，将使超级计算机设施的并行能力的使用成为必要。该研究的完成将大大增加对自然发生的各种尺度干扰在形成模式过程动力学中的作用的理解，为进一步对不同类型的草地生态系统进行比较调查提供建模框架，并为比较草原与森林提供坚实的基础。并举例说明超级计算机设施在增进对生态的认识方面的重要性。

The objectives of the proposed research are (1) to develop a systematic, computerized procedure to detect the structural patterns in the grassland landscape of Jasper Ridge, California, based primarily on the 2-Dimensional Net Function Interpolation Theory; (2) to construct and validate a hierarchically structured, spatially explicit, stochastic simulation model of pattern-process dynamics, based mainly on data from the Jasper Ridge grassland ecosystem. The model will take into consideration processes involving small-scale/high frequency, middle-scale/low frequency, and large-scale/very low frequency phenomena. Stochasticity in each process will have to be incorporated. Large-scale phenomena may include weather fluctuations (e.g., precipitation, temperature), wind/water erosion, prolonged drought, and irregularly timed, fires. Small- scale phenomena, on the other hand, may involve physical environmental patchiness and activities of mound-building animals such as gophers and ants. Overall, the model will incorporate demographics of small-scaled disturbances in grasslands, monitor the temporal changes of patches, and trace aging and extinction of the patches at various scales. The spatially explicit, multiple-scale pattern induction model takes a bottom-up approach, deriving landscape-level dynamics of the grassland from that of hierarchically decomposable lower levels. The computational demand in spatial and temporal statistical analyses of patterns, and the sensitivity anaLysis and Monte Carlo simulation of the model, will make desirable and imperative the use of the parallel capabilities of the supercomputer facilities. The completion of the proposed study will add substantially to understanding of the role of naturally occurring disturbances at various scales in creating the pattern-process dynamics, provide a modeling framework for further comparative investigation on different types o grassland ecosystems, serve as a strong basis for comparing grasslands with forests, and exemplify the significance of supercomputer facilities in enhancing ecological understanding.